OSU Libraries and Research Dataset
Curation: A Beginning
Research & Innovative Services Report 4
February 12, 2009
Bonnie Avery
May Chau
Ruth Vondracek
Andrea Wirth
Table of Contents
Table of Contents ............................................................................................................................................... 1 Executive Summary ............................................................................................................................................ 3 Research Sponsor........................................................................................................................ 3 Project Participants .................................................................................................................... 3 Project Scope and Assumptions .................................................................................................. 3 Process ........................................................................................................................................ 4 Summary of Results ..................................................................................................................... 4 Recommendations ....................................................................................................................... 4 Full Report ......................................................................................................................................................... 6 Introduction ....................................................................................................................................................... 6 Sponsor: ...................................................................................................................................... 6 Project participants: .................................................................................................................. 6 Purpose of investigation: ............................................................................................................ 6 Methodology ..................................................................................................................................................... 6 Background ........................................................................................................................................................ 7 OSU Libraries Subject Librarian Expertise .............................................................................. 7 Content: Aggregating geospatial datasets about Oregon .......................................................... 8 Content: Dataset repositories and curation for Oregon Explorer partners ............................... 8 Storage and access: OSU spatial dataset repository and ScholarsArchive ............................... 9 Literature Review .............................................................................................................................................. 11 Literature Review Results ......................................................................................................... 12 Survey of Issues ................................................................................................................................................ 13 Survey results ............................................................................................................................ 15 Conclusion ........................................................................................................................................................ 18 Cited References ............................................................................................................................................... 20 Appendix A: King & Wirth Reports on GIS & Data Management Librarian ......................................................... 23 Appendix B: Dataset Management and Curation Survey .................................................................................... 33 Appendix C: List of Universities ......................................................................................................................... 38 RIS Report. No. 4: OSU Libraries and Research Dataset Curation 2
Executive Summary
Research Statement
Identify strategies and resources which have proved successful at other libraries where programs
for campus-wide dataset curation are in place. Articulate common "problems" that are
encountered by the implementers of such programs. Make recommendations for further
investigation and an "upgrade" or enhancement of service based on success at other libraries.
Research Sponsor
Faye Chadwell
Project Participants
Bonnie Avery, May Chau, Ruth Vondracek, Andrea Wirth
Project Scope and Assumptions
This research will:
• focus on the management and curation of born digital data sets either active or inactive,
with particular emphasis on GIS data;
• identify strategies which have proved successful at other libraries where programs for
campus-wide dataset curation are in place;
• address commonly articulated “problems” that are encountered by the implementers of
such programs;
• include resource allocation questions when surveying other libraries, such as positions
associated, time it takes, training, hardware, software, maintenance, preservation.
• discuss curation capabilities/resources currently in place at OSU Libraries
• be “informed” by prior efforts to define issues related to campus dataset curation
(primarily focused on particularly spatial datasets) and by the survey of “data librarians”
positions undertake by Andrea Wirth and Valery King.
• This research project will not:
• explore the curation and management of large scale datasets generated by
supercomputers, such as those produced by USGS, NOAA, and NASA;
• not aimed at addressing the whole of the OSU research dataset curation issues on
campus;
• not recommend whether or not to put a program in place at OSU Libraries.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 3 of 41 Process
The research group of Bonnie Avery, May Chau, Ruth Vondracek, Andrea Wirth reviewed past
communications concerning creation of a data center on the OSU campus. We conducted a
literature review and web scan in order to identify libraries with data curation and management
programs in place and to identify articles that dealt with data curation and management issues.
We then searched the libraries’ websites for further documentation. Using mind mapping
techniques we defined gaps and common issues in the information found during our research and
used that information to develop survey questions. We surveyed sixteen libraries.
Summary of Results
The literature and web searches, and the survey results revealed that no one existing model
matches OSU Libraries’ situation because these universities do not have the same infrastructure
and they have greater resources. Each of these universities do provide solutions, tools, or
processes that have implications for OSU Libraries, such as Oxford’s planning process, Purdue’s
cheat sheet for librarian-researcher interview, and Cornell’s overall’s response to developing
expertise in this area. Also responses to the survey indicate that many libraries, like us, are in the
beginning stages and exploring the feasibility of establishing dataset curation and management
services.
Recommendations
•
•
•
•
•
•
•
The greatest impediment to dataset curation and management has been storage. Storage
solutions (other than the Libraries’ investing in more servers) and costs should be
explored further. Terry Reese has begun this work, but a small task force could be
established. A member of UO’s library staff could be included on this task force.
More investigation should be undertaken in management of dynamic data and datasets
and their use in e-science.
OSU Libraries should lead the effort with its partners to conduct a data inventory on
campus, focusing on GIS data. Brian Westra, UO, has expertise in this area and should
be consulted.
OSU Libraries’ should conduct a survey or interviews with OSU researchers to establish
what is needed or lacking in data curation and management services on campus. This
would include following up on the work done earlier by Cathy Howell’s group.
Archiving has already begun in ScholarsArchive with static datasets. What is needed at
this point is the development of policies and procedures specifically related to datasets.
The Digital Repository Work Group could assign a task force to work on this issue.
The role of subject librarians in acquiring datasets and liaising with faculty should be
defined. This responsibility could reside with either Collections or the Digital Repository
Work Group.
To move forward in the development of services a program of staff development and
training should be developed. This would involve training subject librarians so that they
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 4 of 41 can be conversant enough to liaise with faculty and student researchers. Other selected
staff would require training in how to curate and manage data.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 5 of 41 Full Report
OSU Libraries' and Research Dataset Curation
Introduction
Sponsor: Faye Chadwell
Project participants: Bonnie Avery, May Chau, Ruth Vondracek, Andrea Wirth
Purpose of investigation:
OSU Libraries has considered the issues surrounding the collection, curation and management of
born digital datasets, either dynamic or static, for some time. Questions have arisen about the
level of involvement needed, personnel issues and training, and storage issues and costs. This
research project undertaken by Research & Innovative Services (RIS) sought to identify
strategies and resources which have proved successful at other libraries where programs for
campus-wide dataset curation are in place. In addition we wanted to articulate common
"problems" that are encountered by the implementers of such programs. On the local level we
identified the curation capabilities and resources currently in place at OSU Libraries and
reviewed the history of campus-wide discussions concerning establishing data centers. From this
research we planned to make recommendations for further investigation and an "upgrade" or
enhancement of service based on success at other libraries. For the purposes of this research
project, we determined that curation and management of large scale datasets generated by
supercomputers, such as those produced by USGS, NOAA, and NASA were outside the scope of this
research.
Methodology
To understand the background on OSU Libraries’ interests in dataset curation services we
reviewed past communications concerning creation of a data center on the OSU campus and
talked with library staff associated with ScholarsArchive. We conducted a literature review and
web scan in order to identify libraries with data curation and management programs in place and
to identify articles that dealt with data curation and management issues. We then searched the
libraries’ websites for further documentation. Using mind mapping techniques we defined gaps
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 6 of 41 and common issues in the information found during our research and used that information to
develop survey questions. We surveyed 16 libraries.
Background
Historically, OSU Libraries’ interests related to dataset curation services fall into four somewhat
interrelated areas, all related primarily to spatial datasets:
• Development of OSU Libraries Subject Librarian Expertise;
• Content: Aggregating geospatial datasets about Oregon;
• Dataset repositories and curation for Oregon Explorer partners;
• Storage and access issues, most specifically related to developing an OSU spatial dataset
repository and inclusion of datasets in ScholarsArchive.
OSU Libraries Subject Librarian Expertise
Currently OSU Libraries does not have a data librarian position or anyone specifically trained in
to manage or curate datasets. Sue Kunda, the Digital Production Librarian does have
responsibility for adding static datasets to the ScholarsArchive. Andrea Wirth’s (Geosciences
Librarian) responsibilities do not include data curation or management.
This research project is informed by prior efforts to define issues related to campus dataset
curation, primarily focused on particularly spatial datasets and by the reports on of data
librarianship positions undertake by Andrea Wirth and Valery King. The geosciences/maps
librarian position description at OSU Libraries has evolved overtime to acknowledge the
importance of GIS/spatial datasets and tools. Likewise the government documents librarian has
some responsibility for datasets in CD format from the US Census Bureau and other agencies.
Prior to their general distribution via the web, OSU Libraries developed the Government
Information Sharing Project (GovInfo), a web site that ran from 1995 through 2003. The site
provided access to census, population, housing, agriculture, economics, and education resources.
OSU Libraries terminated the website once the government began to distribute the datasets on
the web.
Until recently, the extent to which GIS knowledge was distributed across campus mitigated the
need for the library to assume a central role in locally produced dataset curation. In the spring of
2007 Andrea Wirth and Valery King prepared a report for the library on the range of services a
Data Services Librarian might offer based on their investigations of peer institutions and the
literature (Appendix A). In their report they summarized this range shown below.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 7 of 41 Wirth (A
Appendix A: pg. 20) lists the known units
u
using GIS
G on the OSU
O
campuss. To date noo
inventoryy has been co
ompleted wiith these units.
Conten
nt: Aggre
egating geospatiall datasets
s about Oregon
O
This secttion includess detailed infformation onn OSU Libraaries interacttions with itss main OSU
departmeental partnerrs and contaccts in attemppting to instittute a data ceenter on the OSU Campuus.
While grrant funding has not beenn secured thiis backgrounnd helps to elucidate
e
OSU
U Libraries’’
relationshhips and its work
w
to datee and concerrns using DS
Space and LibbraryFind foor dataset
curation and manageement.
OSU Libbraries particcipated in plaanning and establishing
e
for “Virtual Oregon”, a data
coordinattion center at
a Oregon State Universiity (OSU) beetween 20000 and 2004. It was
established in order to:
t “(1) archiive environm
mental and other
o
place-bbased data onn Oregon andd
associateed areas; (2) make those data accessiible to a broaad spectrum of agencies and individuuals
via innovvative web in
nterfaces; (3) identify keey data sets that
t are not yet
y availablee and encourrage
their colllection and dissemination
d
n; and (4) faacilitate deveelopment of statewide standards for
1
archivingg, documentiing, and dissseminating data.”
d
Virtual Oregon
O
functtioned as a portal
p
using a “distributedd architecturre that occuppies multiplee
locationss…". The fo
our ‘nodes’ thhat compriseed Virtual Oregon
O
the Department
D
o Geosciencces
of
(College of Science),, the Forestryy Sciences Laboratory
L
(U
USDA Forest Service annd College of
o
Forestry)), the Northw
west Alliancee for Compuutational Science and Enngineering (N
NACSE), and the
Valley Library. NAC
CSE (Northw
west Alliancce for Compuutational Sciience and Enngineering)
f
Tim
m Fiez develloped
discontinnued this firsst generationn Virtual Oreegon site duee to lack of funding.
a second generation site
s that wass the precursor to Oregonn Explorer thhat had a muuch modifiedd
form andd a more limiited partnersship. The Viirtual Oregon site itself is
i now defunnct.
Conten
nt: Dataset reposiitories an
nd curatio
on for Ore
egon Exp
plorer
partne
ers
In additioon to the neeeds identifiedd in the Virttual Oregon discussions,
d
OSU Librarries encounters
dataset cuuration serviices as a neeed among its Oregon Expplorer partneers, agenciess and commuunity
groups, particularly
p
the
t need for novice and/oor expert rettrieval, maniipulation of, if not storagge of,
RIS Reportt. No. 4: OSU LLibraries and R
Research Datasset Curation Page
e 8 of 41 datasets produced outside of OSU. Accessibility of OSU-produced datasets continues to be an
issue as well.
Storage and access: OSU spatial dataset repository and
ScholarsArchive
In the summer of 2006, Kathy Howell, then Co-chair of the Faculty Senate Computing
Resources Committee, initiated a meeting with key campus staff and faculty to discuss the
feasibility of establishing a spatial data repository for OSU and invited the library to participate
in the hopes that the libraries’ DSpace digital repository might prove a suitable platform. (Kathy
Howell email 06/08/06)
By the time the meeting took place the agenda had changed to:
1) Identify the broad needs of the campus related to spatial data, especially as relates to how it is
discovered, stored, accessed, and organized (our “audience” is the internal OSU campus
community, as statewide/government agency needs being met by efforts of INR, Library’s
Explorer Series, in collaboration with State Geospatial Enterprise Office).
2) Develop a general strategy for addressing the identified campus needs. Our efforts will
ultimately be successful if a strategic plan is developed which the university enacts upon.
OSU Libraries’ attendees included: Bonnie Avery, Tim Fiez, and Jeremy Frumkin. Others
attending: Todd Jarvis, Institute for Water & Watersheds; Jimmy Kagan and Kuuipo Walsh,
Institute of Natural Resources (INR); Cherri Pancake and Dylan Keon, NACSE (Northwest
Alliance for Computational Science and Engineering); Sheila Slevin, College of Agriculture
(COA)/Crop&Soils; Chris Romsos and George Taylor, College of Oceanic and Atmospheric
Sciences (COAS); Theresa Valentine and Terralyn Vandetta, College of Forestry (COF); Dawn
Wright, College of Science (COS) Geosciences Department.
After a second meeting the group decided to host three “pilot projects” in the hopes these would
better define overall scope and cost for other units around the university and make a more
realistic case to the university or to funding agencies for long-term support.
Case studies included Forestry, Geosciences, and the Natural Heritage Program. Forestry’s pilot
project, with Theresa Valentine acting as data librarian, was to figure out the best way to map
drives to campus networks, to advertise them and provide web accessible data. Geosciences,
with a graduate student serving as data librarian, focused on trying to make ScholarsArchive
work for spatial data. The Natural Heritage program wanted to find ways to distribute
effectively its archived data as well as non-archived. Kuuipo Walsh served as data librarian.
The group agreed that the “Characteristics of successful pilot projects” would be:
Consultancy with Jeremy Frumkin, Tim Fiez and Terry Reese in order to articulate needs and
identify specialized solutions, such as ScholarsArchive or Geospatial One-Stop.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 9 of 41 Well documented. All procedures and potential pitfalls would be documented in a ‘cookbook’ to
inform protocols and procedures for the future.
•
•
•
•
Create a campus standard for minimum, abstracted metadata format. That is agreement
that everyone must write metadata, everybody must provide metadata in this simple form,
regardless of other methods in use such as FGDC, Dublin Core, etc.
Provide case study scenarios to know how people will use the data. As an example, the
Geosciences pilot tried their solutions out on instructors and students in courses to see if
these courses could access data for classroom use.
Create a methodology that others can use.
Would be funded or has implications for future funding.
At a later meeting the OSU Libraries team shared DSpace’s limitations and why it would not be
the ideal storage repository for large, active datasets. Library representatives shared the
attributes and limitations of DSpace as they related to archived datasets. “For example, DSpace
archives files or set of files for download only; it can provide a persistent link for the file; it
could use any metadata scheme – Dublin core, FGDC, although FGDC was not set up as a
default; there would need to be some work to set up forms in order to handle entry of data and
search on fields; it is good for large files size (2 GB or larger hard to do via web interface); not
good for linking to other places via the metadata, it could for certain types of formats, but that
would take further development. D-Space will cover MS-Word, but not GIS, although this would
need to be policy for future migration. Storage is ultimately an issue with files other than
documents; would require special development and programming for migration or to add full
description and adding forms for retrieval (one time cost.)"
At this time the limitations of the open source LibraryFind tool related to GIS referencing
seemed to place Geospatial One Stop (GOS) Arc9 toolkit in a better position to meet the
immediate need for these case studies. Even with a system that delivers the workflow, archiving,
and persistence components there remains the question about what other features and functions
are needed and where do they reside.
The discussion of dataset audiences and access issues:
• Researchers: Faculty/staff/graduate students need a stable base data layers, 10-m DEM, that
include watershed boundaries, road networks, stream networks, specialized land use, and
hydrogeomorphic boundaries. They would also need a tool to help people automatically connect
to the data and import/export/scale/ it.
• Teaching faculty need to locate data and tools. LibraryFind was developed for broad searches
for undergraduates rather than as a deep searching tool for faculty.
• Building a bridge from LibraryFind to GOS might be possible in the future.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 10 of 41 • Building in campus standards for harvesting of metadata
In February 2007 Tim Fiez, Dawn Wright, and Theresa Valentine submitted a $10,000 Proof-ofConcept Grant Proposal to the Northwest Academic Computing Consortium (NWACC) for a
Spatial Data Mapping and Distribution System: Implications for Oregon "Collaboratories" which
was not funded. During the spring of 2007 the University also relinquished responsibility for
central distribution of software contracts. Significantly the ESRI contract oversight was assumed
by Geosciences on behalf of others on campus. By the fall of 2007, the Campus Spatial Data
Repository Group was on hiatus. By that time library representation on the group was limited to
the technical team from the Oregon Explorer.
This group recently reconvened to determine whether the concept of an OSU Data Center could
be resurrected with a similar proposal that might be a viable candidate for a grant (NSF – Math
and Science Partnership (NSF-MSP) Innovation through Institutional Integration.
Representatives from Linn-Benton Community College also attended this meeting. It is unclear
at this time if a proposal would be appropriate for this particular grant opportunity.
ScholarsArchive is used to store some static datasets. The majority of these are associated with
theses and dissertations. Dataset inclusion in ScholarsArchive is determined on a cases-by-case
basis. To use the datasets users must download the data and provide their own software to
manipulate them.
It should be noted that Terry Reese recently submitted a grant proposal that if accepted will
impact the Libraries’ involvement with dataset management and curation. This is a MIT-led
proposal to the NSF Office of Cyberinfrastructure’s DataNet Program on the MIT DataSpace
project, federated data curation system. Included in that proposal is a request for both a research
data analyst and a systems analyst.
While this background informs OSU Libraries that there is a longstanding need for a spatial, if
not also non-spatial, dataset repository for the campus, it is not clear that “participating in the
discussion” has been a sufficient tactic for contributing to the solution. Rather, an articulation of
how the current library expertise and infrastructure can be the solution to a piece of the campus
puzzle is the tact we have taken in this investigation.
For that reason we looked to cast a broad literature informed net to find libraries that seem to be
a step ahead of us.
Literature Review
The rationale for our literature review was threefold.
• Find institution(s) that had dealt successfully with its need for dataset storage and curation.
• Identify additional issues concerning dataset curation we had not considered.
• Identify good candidate institutions for further investigation.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 11 of 41 We searched LibraryLit, ACM Digital Library and Google for articles and documents. Where it
seemed appropriate we used the Web of Science and Google Scholar to check the articles that
were found as cited references. Once libraries were identified we searched their websites for
further documentation.
The scope of this literature review is focused on identifying strategies and resources other
libraries used to build campus-wide dataset curation programs. Since this concept is relatively
new to the library community, literature about its development is scarce. A few -libraries in the
world had established functional working groups for data curation programs, their scopes are
broader than the RIS intended. These libraries collect a variety of data including text, statistics,
images, objects etc, while Oregon State University Libraries (OSU Libraries) dataset curation
project team is more interested in specific datasets (e.g. GIS). Nevertheless, publications
produced by other libraries, including Cornell, Monash and Oxford provide valuable
information; based on their experiences and the well thought out solutions they offered.
Literature Review Results
The CUL Data Working Group from the Cornell University Libraries (CUL) published a white
paper Digital Research Data Curation: Overview of Issues, Current Activities, and
Opportunities, to furnish “an overview of the current landscape and issues surrounding data
curation, and includes recommendations for CUL in this area”2 This report is one of the most
thorough reports available in terms of their environmental scanning, identifying issues and
recommendations. It is a ‘must read’; several of their recommendations seem applicable to OSU.
Monash University, Australia, communicated that the idea of the one single repository approach
is not sufficient to support the entire research cycle. The report introduces the concepts of
curation continuum, domains division and curation boundary3. University of Oxford reports a
comprehensive investigative plan. Steps include stating of objectives, presenting overall
approach, planned project activities, expected outcomes and analysis of stakeholder and risk.4
This particular document may be of use as OSU Libraries pursues further development in data
curation and management services.
Purdue University libraries also sees the need to change and thus created the Distributed Data
Curation Center (D2C2) to “serve as a mechanism to bring researchers together to investigate
ways in which optimal dataset management can be achieved at Purdue and throughout the
research world”.5 In addition, a review specifically on literature focused on geospatial repository
construction and design is also available.6
The rest of the selected literature targets many important issues such as data management,
collaboration, user requirement analysis, staff development and jargon associated with data
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 12 of 41 management. In the literature, data management is a frequent topic, which includes data
collections at the research level7, 8, accessibility4, 6, 7, 9, preservation 9, metadata standards 9, 10, 15
usability (use and reuse) of data 11, 12, policy13, 14 and quality of data.9
Collaboration7 is not limited among researchers; the librarians’ role 9, 15 in data curation has been
discussed. For example, Purdue produced a cheat sheet for librarian-researcher interview for data
curation.16 Another piece of information needed to establish a successful data curation program
is user requirement analysis. The Digital Curation Center in the United Kingdom published a
user requirement analysis report, in which users are “discussed from the point of view of their
roles in functional relation to research data, and also from the point of view of significant
functions of organisational entities.”17
The literature also provides insights on staff development in building data curation programs. It
covers the areas of current practices and training need of research staff.16, the need to provide
“both institutional capacity and appropriately qualified individuals”18 is also identified.
Furthermore, jargons6, 14, in data curation programs are at times obscure and need to be further
defined (e.g. archives4, repository6, Cyberscholarship11).
Survey of Issues
Informed by the literature review we brainstormed questions for further inquiry. Using mind
mapping we categorized these questions and identified relational linkages (primarily,
communication but funding and expertise also apply). This served as a means of organizing our
survey and managing redundancy in our questions.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 13 of 41 We inquired about four aspects of dataset curation/preservation activities. These were:
• User services
• Partners and collaborators
• Technology
• Staffing levels and organization/management for this activity
We elected to use SurveyMonkey for our survey tool. The survey questions can be found in
Appendix B.
Survey participant list:
Prior to culling, we had identified a list of 39 institutions. That list is available in Appendix C.
To cull this list, we investigated their websites looking for:
• The presence of an institutional repository;
• Evidence of archived datasets and/or the inclusion of datasets in their IR guidelines for
collection scope; and
• Other evidence to indicate that planning for dataset acquisition was in process.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 14 of 41 This resulted in the following list of 16 candidate institutions for which we then located contact
information. In November, 2008 we sent an email to each contact that included a link to the
survey.
• Australian National University
• Cornell University
• MIT
• Purdue
• Rutgers
• Stanford
• University of Illinois
• University of Kansas
• University of Oregon
• University of Utah
• Utah [statewide] Digital Repository
• University of Washington
• California Institute of Technology
• Monash University Library (Australia)
• Scripps Institute of Oceanography
• Woods Hole Oceanographic Institute
Two other requests for information appeared on the SPARC listserv related to this topic as we
were finalizing our survey. One was an open ended survey from Gabrielle Wong of the Hong
Kong University of Science and Technology Library. We communicated with Ms. Wong and
agreed to share our results. Then on November 13, 2008 SPARC served as the vehicle for a
survey request from the Digital Curation Centre seeking input for a major international survey on
digital preservation.
Whether these “competing” requests for information had the effect of adding survey fatigue for
those presented with our survey is unknown. For whatever reason, we did not receive the
volume of results we’d hoped, though we can benefit from some of the comments make in
response to the survey from the Hong Kong University of Science and Technology Librarian. To
date replies to this query have been received from MIT, UI/UC, Optical Society of America, and
the Smithsonian. Survey results
We sent invitations to complete our survey to 16 institutions. Eight of these initiated the survey
and three completed it. Of the five minimal responders, four indicated they were in the
“planning stages” while one respondent characterized the extent of data preservation was
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 15 of 41 “extensive” but provided no further detail. We received phone calls and emails from staff at
three institutions, who felt that they were not far enough along to answer the questions; these
individuals were asked to complete what they could.
University of Illinois and Cornell provided the most information and indicated that follow-up
inquiries were fine. The other respondents indicated no interest in providing follow-up
information and/or provided no contact information. Still since the University of Illinois
respondent indicates that they are “in the planning stages” and Cornell, that they are “well
underway”, their responses proved useful. To this we elected to add the response from MIT to
the SPARC LISTserv/Hong Kong University of Science and Technology request for feedback on
collection research datasets.
Respondents indicate that they are offering their curation/preservation services to the entire
campus and Cornell indicates that they are also working with a “handful” of state and local
agencies. The services offered are long term storage, metadata assignment (assistance with
identifying what metadata is needed), and in the case of Cornell: location aids for users, dataset
manipulation software, collaborative online workspace (via Wiki, and DataStaR).
Guidelines for dataset collecting scope are most developed at Cornell and include:
• eCommons (quotes from survey)
• GIS Data Repository
• DataStaR
Issues related to rights management include further investigation into who actually owns the
dataset (a case of this nature was mentioned by the UI respondent) and concerns researchers may
related to “misuse” of their data (providing sample rights statements for them to modify).
Cornell’s dataset repository efforts began with an NSF funded pilot project and continue with
collaboration between the library and the Center for Advanced Computing on their campus.
Dataset deposition has been voluntary in all cases. Early adopters on campus were in plant
science, anthropology and animal science at UI. They credit some of the interest in “sharing”
with the fact that UI is a land grant institution with an Extension service. Cornell notes that early
adopters tend to be from one of two communities:
• Those who rely on data from a shared facility or instrument (e.g. physics and astronomy).
• Those who work in a field where comparative or long-term studies make shared data valuable
(e.g. ecology, genomics).
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 16 of 41 MIT noted that, “The researchers here haven't been shy about sharing their data in many cases.
Those who are can talk to us about limiting access or setting an embargo period (that goes
against our standard policy, but we can make exceptions if they're justified). Luckily copyright
hasn't really been an issue since in the U.S. you can't copyright data, which does lead some
researchers to want to keep their data locked up so that's a policy issue you'll have to work
through for yourselves.”
While the institutional repository serves as a home for datasets for all respondents, some datasets
are also stored on departmental servers. In some cases researchers are using commercial services
as well. Datasets associated with electronic theses and dissertations (ETDs) and/or current
student/faculty research are collected by all respondents and include:
• archival (closed) datasets,
• versioned datasets, and
• active/dynamic datasets (Cornell).
While UI integrates these services under its IDEALS rubric, Cornell indicates that integration of
non-IR aspects of dataset curation are somewhat linked to the expertise and promotion by the
librarian. To date, respondents indicate that depositing of datasets is most often done by library
staff, though the dataset provider is authorized to do this task.
Cornell relates that problems of interoperability they noted include use of acceptable file formats
and metadata. Their data librarian works with the researcher to reformat datasets, correct
metadata, and be knowledgeable as to best practices.
Google-like searching and the institutional repository descriptions remain the primary means of
third party discovery of datasets. Cornell experimented with extracting GIS metadata and
converting it to MARC records but discontinued as it required too much work for the impact.
They found that the catalog was not a path for dataset discovery.
The role of the library is quite varied:
• UI indicates that librarians work with faculty to get appropriate sets of material to deposit into
the IR including the data, provenance and protocol information, and other publications that have
resulted from the data. They indicate that it is not clear what is happening in other units on
campus related to dataset curation.
• Cornell indicates that neither the library nor other units have formal responsibility for dataset
curation but that the library is active in this area as are some other units.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 17 of 41 IT staff have responsibility for dataset storage while dataset description and user services area is
a collaborative effort between the library and dataset provider. Areas of needed expertise noted
are: subject knowledge, knowledge of metadata standards, knowledge of the repository set up
and management, current best practices for dataset preservation, and research computing.
Consultation on what datasets are important to share and expertise in specific metadata standards
were noted as on-going areas of need.
Current levels of staffing also vary. At UI the IR staff can also depend on “portions of subject
liaisons based on FTE.” At Cornell, three librarians devote a significant amount of their time to
curation efforts. In addition, six to eight IT staff members spend a portion of their time
supporting dataset curation services (including IR services). In both instances, old positions
were “repurposed” to accommodate the need for staff. This repurposing includes a redefining of
subject librarians’ positions as well as folding in “datasets” within the collecting scope of the IR
and its staff.
Respondents consider their efforts to be “successful” and that dataset curation has been well
received by faculty. However, they indicate that it remains to be seen if their efforts are scalable
and sustainable.
Conclusion
Cornell Libraries’ white paper describes reasons for pursuing data curation and management in
libraries succinctly.
There are three primary (and related) motivations for developing a robust data
curation infrastructure: enabling new discoveries by exposing data for use in datadriven research, ensuring access to and preservation of scholarly output, and
meeting existing or forthcoming requirements of funding agencies or institutions
regarding data management, retention, and access. Libraries have demonstrated
expertise in several areas that could be productively applied to the practice of data
curation, and in some cases, cyberinfrastructure development.19
OSU Libraries’ has developed strong relationships with other campus entities that may lend
themselves to developing on-going partnerships if we decided to pursue developing a robust
suite of data curation and management services. Our current server infrastructure does not
support storage of a large volume of datasets and we do not provide tools to manipulate the data.
For that reason, we have relied on a distributed data model that continues to be a viable option.
Other solutions for storing datasets are available on campus, such as NACSE and externally,
through commercial means, but need to be explored further.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 18 of 41 This report represents a preliminary investigation into the issues surrounding dataset curation and
management and libraries involvement. We found that the most complete information about
planning and implementing a program came from Cornell, Oxford and Purdue Universities.
The drawback at this point is that none of these models match OSU Libraries’ situation because
these universities do not have the same infrastructure and they have greater resources. Each of
these universities do provide solutions, tools, or processes that have implications for OSU
Libraries, such as Oxford’s planning process, Purdue’s cheat sheet for librarian-researcher
interview, and Cornell’s overall’s response to developing expertise in this area. While we may
not be able to follow a specific model, we could certainly adapt some of these practices in
creating our own model.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 19 of 41 Cited References
1
Keon, D., Pancake, C., and Wright, D. 2002. Virtual Oregon: seamless access to distributed
environmental information. In Proceedings of the 2nd ACM/IEEE-CS Joint Conference on Digital
Libraries (Portland, Oregon, USA, July 14 - 18, 2002). JCDL '02. ACM, New York, NY, 387-387. DOI=
http://doi.acm.org/10.1145/544220.544334
This article provides an excellent overview of the Virtual Oregon project and the motivation behind its development.
Provides a highly technical view of the infrastructure and software used to develop and implement the site.
2
Steinhart, G. et al. (2008). “Digital Research Data Curation: Overview of Issues, Current Activities, and
Opportunities for the Cornell University Library” (http://hdl.handle.net/1813/10903)
3
Treloar, Andrew and Cathrine Harboe-Ree (2008). “Data management and the curation continuum: how
the Monash experience is informing repository relationships”.
(http://www.valaconf.org.au/vala2008/papers2008/111_Treloar_Final.pdf)
4
Martinez Uribe, Luis (2008). “Scoping digital repositories services for research data management: a
project of the Office of the director of IT”. Oxford e-Research Centre, University of Oxford.
(uk/odit/projects/digtialrepository/docs/DigRepoProjectPlan.pdf) February 27, 2008 ver. 2.2
5
Mullins, J.L. (2007) “Enabling international access to scientific data sets: creation for the distributed
data curation center”. (http://www.iatul.org/doclibrary/public/Conf_Proceedings/2007/Mullins_J_full.pdf)
Discusses how Purdue University Libraries created the Distributed Data Curation Center (D2C2) in response to the
needs of researchers and requirements for NSF funding.
6
Bose, Rajendra (2006). “Geospatial repository literature review”.
(http://homepages.inf.ed.ac.uk/rbose/pubs/GRADE_RepReviewFinal_rkb.pdf)
As the title indicates this article is about geospatial data repositories. The purpose of the article is to communicate
the information gathered during the planning phase of the EDINA GRADE (http://edina.ac.uk/projects/grade/)
project. After a review of what it means to be a repository or archive, the author provides a brief synopsis of many
existing geospatial repositories often including the scope of the collections and the types and sources of data.
Metadata standards are also reviewed. The article is good summary of geospatial data repositories, but I think the
most helpful part of the article is in the summary where the author describes how the GRADE group is looking at
managing datasets - both depositing and accessing data and issues of curation and metadata assignment (curation
in their case means editorial review of the deposited datasets).
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 20 of 41 7
Palmer, Carole L., Melissa H. Cragin, P. Bryan Heidorn, and Linda C. Smith. “Data curation for the
long tail of science: the case of environmental sciences”. (n.d.)
(http://209.85.173.132/search?q=cache:L3Jz78dGKY4J:https://apps.lis.uiuc.edu/wiki/download/attachme
nts/32666/Palmer_DCC2007.rtf%3Fversion%3D1+palmer+carole+data+curation+evironmental+science
&hl=en&ct=clnk&cd=8&gl=u)
The authors are developing a plan to manage long tail research data generated in an interdisciplinary field
(Environmental Sciences) in order to help address how they might "unleash the market potential of these data
collections and lower the barrier to access and reuse" (p. 2). They also address their project in terms of how a
"coordinated or consortial" (p. 2) Institutional Repository might be a good economically viable and scalable
solution to managing these smaller datasets.
The remainder of the article describes the research study - how the authors are developing it at the time it was
written. They provide a list of their research questions such as "how can research level collections best share and
exchange data with existing resource and reference level collections" (p.3). Following that is a description of their
survey and other data-gathering methods.
8
Arms, William Y. and Ronald L. Larsen. (September 12, 2007) The future of scholarly communication:
building the infrastructure of cyberscholarship. Report of a workshop held in Phoenix, Arizona, April 1719, 2007.” NSF and the Joint Information Systems Committee.
(http://www.sis.pitt.edu/~repwkshop/NSF-JISC-report.pdf).
9
Gold, Anne (2007)”Cyberinfrastructure, data and libraries; Part 1”. DLib Magazine September/October
2007 (http://www.dlib.org/dlib/september07/gold/09gold-pt1.html) and Gold, Anne (2007) and
”Cyberinfrastructure, data and libraries; Part 2”. DLib Magazine September/October 2007
(http://www.dlib.org/dlib/september07/gold/09gold-pt2.html)
10
Henty, Margaret, Belinda Weaver, Stephanie Bradbury and Simon Porter. (2008) “Investigating data
management practices in Australian Universities”. APSR Australian Partnership for Sustainable
Repositories. July 2008. http://www.apsr.edu.au/orca/investigating_data_mangement.pdf
The goal of this survey is to find out what the current practices and training need for research staff. Their target
audience includes academic staff, postgraduate students, emeritus/adjunct and others. It tells us that what
percentage the respondents are generating digital data. What are the types, sizes are being generated. How long the
data will maintains its value, what is the software retention etc. In addition, equipment required in e-research is
also mentioned, ownership etc. One of the interesting things they found out about training is that staff wants
advices on 3 things. (1) Advice on data management plan, digitization, data exit plan for retiring faculty. They also
talked about the interest in "tiers of access" with different privileges granted at different levels.
11
Research information Network (June 2008)” To share or not to share: publication and quality assurance
of research data output. Main Report.” Research Information Network in association with JISC Natural
Environment Research Council(http://www.rin.ac.uk/files/Data%20publication%20report,%20main%20%20final.pdf)
12
Rusbridge, Chris. 2007. “Create, curate, re-use, the expanding life course of digital research data”.
EDUCAUSE Australasia 2007. (http://www.era.lib.ed.ac.uk/bitstream/1842/1731/1/Rusbridge.pdf)
13
Soldi, Miguel (2006). “Safeguarding research data policy and implementation challenges (University of
Texas System). http://connect.educause.edu/Library/Abstract/Safeguarding ResearchDataP/43896
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 21 of 41 14
Carpenter, Leona (2004) “Taxonomy of digital curation users: part of the digital curation center user
requirement analysis.” (http://www.dcc.ac.uk/docs/Taxonomy-dc-users.pdf)
Detail on digital curation users, their relationship to each other. They are categorized by (1) roles in functional
relation to data, (2) significant functions of organizational entity.
15
Pritchard, Sarah M. Smiti, Anand, and Larry Carver. (2005) “Informatics and Knowledge Management
for Faculty Research Data”.
http://connect.educause.edu/Library/ECAR/InformaticsandKnowledgeMa/40109
This Educause Research Bulletin describes an investigation of the needs of faculty researchers in the areas of
informatics on the UCSB campus. The study was funded by the Andrew W. Mellon Foundation.
The study primarily describes the findings from interviews with faculty members who met the criteria of being
technologically innovative and had data-intensive research systems (p. 4). The areas addressed included data
organization, storage, long-term preservation, tools and technique development, metadata application, digital rights
management, and data sharing and system cooperation.
The authors found that lack of time or incentive to seek better methods for data organization, storage, and
preservation were at the forefront. Some felt their current, if informal, systems were "good enough" (p. 6). Faculty
are also heavily influence by what is required from their funding agencies for data preservation. Time was again an
issue in the area of metadata use - even the faculty that were well versed in metadata standards for their discipline
did not always see it as a necessary component of their research.
16
Witt, Michael and Jake R. Carlson. “Conducting a Data Interview.” Purdue Libraries, Libraries
Research Publications (2007).
(http://docs.lib.purdue.edu/cgi/viewcontent.cgi?article=1092&context=lib_research&nbsp)
17
Carpenter, Leona (2004) “Taxonomy of digital curation users: part of the digital curation center user
requirement analysis.” (http://www.dcc.ac.uk/docs/Taxonomy-dc-users.pdf)
18
Henty, Margaret.( 2008) “Developing the Capability and Skills to Support eResearch.”Ariadne. No.#55
April,2008. (http://www.ariadne.ac.uk/issue55/henty/)
19
Steinhart, G. et al. (2008). “Digital Research Data Curation: Overview of Issues, Current Activities, and
Opportunities for the Cornell University Library” (http://hdl.handle.net/1813/10903). pg. 4.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 22 of 41 Appendix A: King & Wirth Reports on GIS & Data
Management Librarian
GIS Librarianship @ OSU
Prepared by Andrea Wirth for Ruth Vondracek
6/2007
Geographic Information Systems (GIS) Librarianship can mean many different things. Some
factors that affect the role that librarians have in GIS services include:
•
•
•
•
Library collections and technical support for GIS users
Librarian knowledge and skills in GIS
Marketing/making known library GIS services
Other sources of support for GIS activities on campus
Library collections and technology support for GIS users
Data sets, data sets, data sets. All areas reviewed for this summary, including a very brief review
of the published literature, a survey conducted by the libraries/Geosciences in 2000/2001, and
comments from librarians, point to patrons with GIS needs typically needing the library to get
help finding, accessing, or storing data sets or digital imagery and maps. Purchasing, linking to,
locating, and accessing data are likely viewed from a reference or consulting services standpoint.
Storing, preserving, and “cataloging” data could involve Library Technology or Technical
Services on a greater scale.
A GIS Librarian would need to:
•
know where to obtain geospatial data sets and be able to respond to requests for specific
data
•
know how to incorporate data into collection development activities
•
obtain local data or direct users to data already owned or created by other members of
campus (if not hosted by the library)
•
solicit data for inclusion in the library’s collection (if the library chooses to include this
aspect)
•
maintain an up-to-date collection of books, serials, maps and other media that are useful
to the GIS certificate programs (professional, undergraduate, and graduate levels) and other units
using GIS (see list at end of this document)
Providing technological support for GIS activities is another area that could be pursued, such as:
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 23 of 41 •
providing workstations that could include the following (modeled after the University of
Kansas Libraries)
ƒ software (ESRI products, Photoshop, and potentially statistical software)
ƒ digital manuals to the software
ƒ data sets included with ESRI products
ƒ local high-use data sets (including aerial photos)
•
providing scanning, printing (high quality, large formats) in the library
•
helping users troubleshoot data downloading
•
software support and distribution
Librarian knowledge and skills in GIS
At minimum, a GIS Librarian should be trained in GIS theory and the ability to find appropriate
data sets for patrons, provide instruction sessions for GIS courses (where finding resources and
introducing library technology, etc. could be addressed). A librarian with more developed GIS
skills could work directly as a GIS consultant and assist in the manipulation of data and
production of end products.
Cooperation between the GIS librarian, map services staff, data services librarian (if separate and
if the position exists), and the government information librarian is essential (as described by KU
Libraries).
Reference
The GIS Librarian would need to be able to respond to a variety of types of reference questions.
A list of reference questions that represent the types of questions KU Libraries received is
included in Hauser’s article. The Librarians here at OSU report that finding data and maps
(digital and print) are the primary types of questions encountered.
If the library aims for a high level of technology support of GIS, the Librarian may also need to
know how to troubleshoot software problems, distribute software, and provide assistance in data
manipulation and map creation (as Scott W had previously done).
Instruction
There are some creative examples of GIS instruction at
http://mapzlibrarian.blogspot.com/2006/03/kolb-learning-inventory-gis.html where the librarian
at University of Texas at Arlington makes GIS instruction interesting by teaching GIS using
“real world” problem solving activities. GIS instruction can also take a much more traditional
approach as well and focus more on spatial data access and literacy and the software used to
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 24 of 41 work with data. There is also interest from the Geosciences for including Oregon Explorer in the
GIS curriculum – both for it’s use as a GIS application and for the administrative technology
behind it.
Marketing/making known library GIS services
If the library begins to offer support of GIS activities on campus, there would be many
opportunities for working across disciplines to define the services more accurately and find a
niche for the library to fill. The OSU GIS community appears to be a fairly open and dedicated
group that already works beyond unit boundaries. Adding the library to this group would not
likely be difficult and would be a good inroad to further assessment of the GIS service needs.
Other sources of support for GIS activities on campus
OSU currently has some level of technical support for GIS users in specific communities. This
would affect where the library needs to pool its resources.
There is currently (semi-official) ESRI software technical support on campus for the following
groups:
•
Forestry community (College of Forestry and Forest Service)
•
College of Agriculture and Crop and Soil Science
•
Geosciences, COAS, and Engineering
Many units that use GIS are not represented in these supported areas above (see list attached at
the end of this document). This could be an opportunity for library services to focus on these
unsupported departments, but this may take a more formal assessment to determine the need with
certainty. As it is primarily the social sciences that are not known to be supported internally, this
may mesh well with the discussion for data services support.
Sources consulted:
In addition to comments from OSU Librarians, a few GIS Librarians from other universities, GIS
Librarian position announcements, and an informal conversation with Dawn Wright
(Geosciences) the following resources were used to gain more insight into how GIS services
could be designed at OSU Libraries.
*Hauser, R. (2006). Building a library GIS service from the ground up. Library Trends, 55(2),
315-326.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 25 of 41 *Stoltenberg, J. & Parrish, A. (2006). Geographic information systems and libraries. Library
Trends, 55(2), 217-360.
Virtual Oregon survey. (2002). http://virtual-oregon.nacse.org/people/index.html
*These both come from the Fall 2006 issue of Library Trends that was devoted to the subject of
GIS and libraries.
List of known units using GIS on the OSU campus (provided by Dawn Wright - Geosciences):
College of Ag
Ag and Resource Economics
Crop and Soil Science
Fisheries and Wildlife
Rangeland Ecology & Mgmt
College of Forestry
Forest Engineering
Forest Science
Forest Resources
HJ Andrews
College of Science
Botany and Plant Path.
Environmental Sciences
Geosciences
Horticulture
Statistics
Zoology & PISCO
College of Liberal Arts
Anthropology
Political Science
Sociology
College of Health & Human Sciences
Public Health
COAS
Oregon Climate Service and Oregon Spatial Climate Analysis Service
Marine Geology & Geophysics
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 26 of 41 Marine Resource Management Program
College of Engineering
Biological and Ecological Engr
Civil Engr
EECS
NACSE
Valley Library, especially Digital Collections and the INR
OSU Extension
Columbia Plateau Conservation Research Center
Klamath Experiment Station
Northern Willamette Research & Extension Center
HMSC
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 27 of 41 Data Services Librarianship @ OSU
Prepared by Andrea Wirth and Valery King with assistance from Michael Baird for Ruth
Vondracek
6/2007
Research in a wide variety of disciplines has increasingly involved the collecting and
manipulation of sets of digital data. Students and other researchers need to access, manipulate
and analyze data sets, and frequently have questions about it. Our users very often need
assistance with using data sets that the library receives from the state and federal government, as
well as the variety of social sciences data that OSU users download from ICPSR. OSU
researchers need to archive the data sets they produce, and are increasingly requesting that the
library acquire more data sets from other sources. We have become ever more aware that there
are some gaps in our collective knowledge and in our services that could be addressed by hiring
a Data Services Librarian.
Defining a Data Services Librarian for OSU depends on the level of support that the library
wants to pursue as well as whether the duties of the position will be combined with other
developing services (such as Geographic Information System services). Below is the spectrum of
data services as summarized from Jim Jacobs’ talk earlier this year at Willamette. This visual
provides the most succinct depiction (that we have found) of what data services can mean at
OSU Libraries. Though the spectrum was designed with social science data services in mind, it
could apply to geographic information systems (GIS) services as well.
Selecting
data
Acquiring
access to
data
Acquiring
data
Organizing
data
Preserving
data
Data
services
(help with
analysis)
Increasing service level---------------------------------------------------------------------------------------->
There is often overlap between numerical data services and GIS services in library literature,
position descriptions, and existing or developing library data services. Two of the position
description reviewed for this summary included a major GIS component: Geospatial Data
Librarian at Emory (http://web.library.emory.edu/services/hr/geospatial_data.html) and the Map
and Data Services Librarian at the University of Illinois - Chicago
(http://www.uic.edu/depts/lib/admin/personnel/map_lib.pdf). The announcements for the
positions at the University of Southern California
(http://www.usc.edu/libraries/jobs/librarians/documents/223DSocSciData.pdf) and at Notre
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 28 of 41 Dame (http://www.library.yale.edu/~llicense/ListArchives/0010/msg00015.html) also explicitly
list working with the Government Documents librarian to facilitate user access to data received
from the government, notably the U.S. Bureau of the Census.
One noticeable feature of these position descriptions is the need for the librarian to work in an
interdisciplinary capacity. While many required skills listed in position announcements for data
or digital librarians are similar to those currently required of subject librarians--collection
development, instruction, reference, liaison and so on—data services librarians do not work
within a single subject area. Whether the focus is numerical data or GIS services (or both), the
librarian will likely assist users from across the campus and may be more of an expert in
quantitative data analysis and/or GIS than an information specialist in any particular subject area.
Skills and duties of a Data Services Librarian:
•
Knowledge of statistical processes and terminology including understanding how users
(individually and generally) want to use data
•
Knowledge of statistical software (SAS, SPSS, STATA are often listed in the position
descriptions we reviewed; SAS, S-PLUS, and R are featured prominently on OSU’s Statistics
department website)
•
Ability to work with data files (downloading data and troubleshooting format issues for
successful use with statistical software)
•
Education or background in the social sciences
•
Collecting (either through purchase or providing access) data sets
•
Creating and maintaining a user interface that provides access to data
o
Example: University of Connecticut Libraries Social Science and Geospatial Data
Services (SSGDS) http://www.lib.uconn.edu/ssgs/index.cfm
o
Example: Emory University Libraries Electronic Data Center
http://einstein.library.emory.edu/
•
Instruction (of library staff and of students and teaching faculty) in the use of library
resources for obtaining data, using/searching metadata, some software instruction.
•
Marketing data services to students and faculty
•
Knowledge of data use on campus (who and which departments are using numerical data)
and coordinating with those departments and other campus resources currently available
(working with subject librarians or directly with departments).
Data Preservation
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 29 of 41 In addition to the above, which could be lumped together as reference and consulting data
services for patrons who need to use data, there is an opportunity for the data services librarian to
act as a Data Archivist (to use Jacob’s terminology) and actively seek to preserve data. This
could include ensuring consistent, perpetual access to purchased or subscribed resources (similar
to other electronic resources issues).
Another direction this could take is the coordination and management of a repository for data
created or owned by campus researchers. This particular aspect of the Data Librarianship
question is important because it has the potential to involve many other sections of the OSU
Libraries, including but not limited to Library Technology and Technical Services but likely
others as well. Creating a secure “place” to store the data as well as developing meaningful
metadata are just two parts to the larger issue of data preservation. In addition, Jeremy Frumkin
(in a brief conversation) pointed out that assessment of campus-wide needs in this area should be
done before designing data preservation or storage services.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 30 of 41 Sources Consulted:
In addition to notes from Jim Jacobs’ presentation at Willamette U in March 2007, the following
resources were consulted.
Bennett, T & Nicholson S. (2004). Interactions between the academic business library and
research data services. Portal, 4(1), 105-22
Gerhan, D. (1999). When Quantitative analysis lies behind a reference question. Reference and
User Services Quarterly, 39(2), 166-76.
Jacobs, J. & Humphrey, C. (2004). Preserving research data. Retrieved May 25, 2007 from
http://3stages.org/jj/w/preserving_research_data.html.
**Read, E. (2007). Data services in academic libraries: Assessing needs and promoting
services. Reference and User Services Quarterly, 46(3), 61-75.
Comment: This article provides a very good analysis of the reference portion of data services
librarianship. It is based on a survey of data use at the University of Tennessee and seems to
address every reference issue encountered in other articles consulted for this write up on data
services. It includes a “Skills and Knowledge” section. It does not address collection
development or data preservation.
Steinhart, G. (2006). Libraries as distributors of geospatial data: Data management policies as
tools for managing partnerships.
Position Announcements consulted:
Geospatial Data Librarian, Emory University
(http://web.library.emory.edu/services/hr/geospatial_data.html) (October 2006)
Map and Data Services Librarian, University of Illinois - Chicago
(http://www.uic.edu/depts/lib/admin/personnel/map_lib.pdf)
Social Sciences Data Librarian, University of Southern California
(http://www.usc.edu/libraries/jobs/librarians/documents/223DSocSciData.pdf)
Data Librarian, University of Notre Dame
(http://www.library.yale.edu/~llicense/ListArchives/0010/msg00015.html) (October 2000)
Social Sciences Data Librarian, Rutgers University Libraries
(http://www.libraries.rutgers.edu/rul/hr/libpersonnel/APP171.pdf) (April 2006)
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 31 of 41 Social Science Data Librarian, Yale University Library (http://data.uwindsor.ca/cgibin/iassist/job.cgi?jobid=7) (May 2005)
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 32 of 41 Appendix B: Dataset Management and Curation Survey
Introduction and Definitions
Thank you for agreeing to complete this survey. The Oregon State University (OSU) Libraries is
in the process of articulating a dataset curation and preservation strategy for our digital
repository, the ScholarsArchive@OSU. To help us guide our efforts we are requesting that you
fill out a questionnaire about your institution’s experience to date. Our specific interests include
user and contributor services, collaboration and partnerships, technology, and impact on staffing
and management.
For clarification:
Datasets of interest are those which are "born-digital" (whether spatial, numeric, etc.) and in
general require some software for analysis and manipulation.
Contributors refers to those individuals or groups who are depositing datasets.
Users refers to those using the datasets once they are stored/perserved.
Audience refers to both of these groups but may in your institution have additional meanings.
Dataset Preservation refers to storage and metadata assignment needed to access datasets.
Dataset Curation refers to contribution, preservation and services needed to provide access (and
If you have questions please feel free to contact ruth.vondracek@oregonstate.edu
Section 1. How would you characterize the extent of dataset preservation/curation by your
library?
Extensive
Well underway
In the early stages
In the planning stages
No plans to preserve datasets at this time
Please comment as needed:
Section 2. User services and dataset curation/preservation.
In this section we are interested in getting an idea of the extent of datasets curation undertaken to
date and services provided to users of these datasets.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 33 of 41 1. To whom do you offer data curation/preservation services?
Faculty
Students
Specific Departments
All Campus
Other (please specify):
2. What preservation/curation services are provided to contributors?
Long term storage
Metadata assignment
Location aids for users
Dataset manipulation software
Collaborative online workspace
Other services
Please elaborate as needed:
3. Do you provide users with copyright information and/or citation guidelines for curated
datasets?
Yes
No
Please describe and/or provide URL if appropriate
4. What is the collection scope for curated datasets? If you can provide a URL with this
information, please do so.
5. Have you encountered issues of rights managements in dataset curation/preservation? If so,
how have these been resolved?
Section 3. Collaboration/Implementation and dataset curation/preservation
In this section we are interested in learning more about partnerships and collaboration between
the library and other units on or off campus that led to dataset curation/preservation.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 34 of 41 1. Did you have a pilot project with a specific group of dataset contributors?
Yes
No
Please comment:
2. What departments, disciplines, other groups have deposited datasets to date or seem most
interested in doing so?
3. Who were the early adopters? If you have thoughts on why, please clarify.
4. Is dataset contribution mandatory or voluntary?
Mandatory
Voluntary
It depends
5. Have you documented or assessed whether posting research datasets has facilitated
scholarship and/or career advancement for the dataset contributor?
Yes
No
Please elaborate as needed:
Section 4. Technology and dataset curation/preservation
In this section we are interested in how you have handled the technology needed to preserve
datasets at your institution and limitations (if any) that available technology places on this
activity.
1. Where are "curated" datasets stored? (Check all that apply).
On library servers.
On institutional repository servers.
On separate computing or information technology department servers.
On commercial service servers.
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 35 of 41 Other
Please comment for clarity:
2. What categories of datasets are accepted for preservation/curation?
Archival (closed)datasets
Versioned datasets
Active/dynamic datasets
Datasets associated with ETDs
Datasets associated with current faculty or student research
Other
Please comment for clarity on "other":
3. Are your dataset preservation/curation services integrated with other library services or
programs (e.g. ETDs, the university IR, digital projects, etc.)?
Yes
No
Please describe:
4. Who actually "deposits" the datasets for curation/preservation in the designated repository?
Library staff.
Dataset creator/provider.
Other.
Please comment for clarity on "other":
5. Have you encountered problems with interoperability (e.g. incompatibility with systems or
formats) reported by either contributors or users?
Yes
No
Please describe how you have/do resolve these issues.
6. What have you done to enable/enhance third party discovery of curated datasets?
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 36 of 41 Section 5. Staffing and Management for Dataset curation/preservation
We are interested in how you have handled issues related to staffing and management for dataset
preservation/curation including expertise and professional development.
1. Please describe the aspects of dataset preservation/curation for which the library is
responsible.
2. Please describe the aspects of dataset preservation/curation which are the domain of other
units on or off campus.
3. Who is responsible for these aspects of dataset curation?
Library staff
IT staff
Dataset provider
Other
Storage:
Library staff
IT staff
Dataset provider
Other
Description:
Library staff
IT staff
Dataset provider
Other
Data set user
Library staff
services:
Please comment for clarity on "other":
IT staff
Dataset provider
Other
4. What expertise was critical in setting up your dataset curation services?
5. In creating this service, what additional expertise (not at your organization) was needed?
6. How many positions and/or what level of staffing is involved with managing and curating
datasets?
7. How did/do you accommodate staffing needs for dataset curation/preservation?
New positions were/will be funded.
Old positions were/will be "repurposed."
8. Has your dataset curation program been successful? How would you characterize it?
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 37 of 41 Appendix C: List of Universities
(Libraries in bold were sent survey)
Library Institution
Institutional Repository
(Y/N)
Datasets in IR (Y/N/?)
Dataset curation or
preservation activity noted?
(Y/N/?)
Australian National
U.Ԝ
Yes
Dataset is a format "type"
in the IR
Yes - surveyԜ
Baylor (GWLA)
Yes
No
No
BYU (GWLA)
Yes
No
No
Colorado State U
(GWLA)
Yes (Ex-Libris)
No
No
CornellԜ
Yes
Emory
Yes -- ETDs only so far
Yes
No
Yes (metadata services)
Data Service Center (mostly
reference services not
preservation)
Iowa State U (GWLA) No
No
No
Johns Hopkins
Yes
No
No
Kansas State U
(GWLA)
Yes
No
MITԜ
No
Yes
2 types of preservation; bit &
functionalԜ
Scholarly publication
repository
Cannot get in full text,
don't have the right.
articles from journals
reference service, preserve
according to standard digital
preservation practices
Yes, Digital repository
Not yet
Yes, excellent planning for
infrastructure, documentation on
preservation and curation
Yes; DSpace
Monash (AU)Ԝ
yes,
ARROW repository
North Carolina State
U
Oxford
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 38 of 41 PurdueԜ
Rice University
(GWLA)
Yes, E-scholar
Yes
E-data is setup, but cannot
Yes
find any data
Has a separate GIS, data
center
some, in Text coding under FAQ
Yes, community repository
Yes, also has a separate
data center
Yes
Yes
No
Yes
Texas A&M (GWLA)Ԝ yes
no
yes
Texas Tech (GWLA)Ԝ yes
no
yes (services to faculty but not
necessarily dataset specific)
U. Arizona (GWLA)
yes
not yet
seems to be in planning stages
this
U Colorado at
Boulder(GWLA)
Not yet but noted in notes for
3/2008 meeting of faculty
No
assembly re: NIH mandate
No
U. Conn
Yes
Seems no
No
U. Hawai'i(GWLA)
Yes (since 2007)
No (not yet?)
Not really
U. IllinoisԜ
yes
YES (some trials data in
Excel and a SAS output
file)
(Yes?)
U. Kansas (GWLA)Ԝ yes
yes (GIS Datasets in a
collection)
(Yes?)
U. Melbourne
yes
not now but mentioned in
yes
collection policy
yes
no but mention of
no but nice chart of preservation
"compilation of university
support provided and other data
data” in content
identification services provided
guidelines
RutgersԜ
StanfordԜ
U. Minn
U. Nebraska (GWLA) Yes
No
No
U. Notre Dame
Yes
No
No
U. Oregon (GWLA)Ԝ Yes
No
Yes?- recently closed a position
announcement for a Data Services
LibrarianԜ
U. So. California
(GWLA)
Yes
No
No
U. Utah (GWLA)Ԝ
Yes
Yes (their IR indicates
Yes if counting the mention of
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 39 of 41 datasets are welcome - did adding datasets to the IR.
not find any however )
U. Washington
(GWLA)Ԝ
Washington State U.
(GWLA)
Washington U. (St.
Louis) (GWLA)
Yale
Yes
Yes, in their Vision 2010 doc
Yes (their IR indicates
they mention progress towards
datasets are welcome - did
this service
not find any however)
Yes
No
Yes
Yes, but pretty limited
No
No
No
No
No
Caltech
Yes
Yes (BLOB)
Yes
Scripps Institute Of
Oceanography (SIO)
Yes
Yes
Yes
Woods Hole
Oceanographic
Institute
Yes
Yes
Yes
RIS Report. No. 4: OSU Libraries and Research Dataset Curation Page 40 of 41